geom_pie()¶

A pie chart is a circular statistical graphic, which is divided into slices to illustrate numerical proportion.

Default presentation

1.4. Use "count2d" statistical transformation

Pie size depending on data
Explode

In [1]:

from lets_plot import *
from lets_plot.mapping import *

LetsPlot.setup_html() 

In [2]:

blank_theme = theme(line=element_blank(), axis=element_blank())

In [3]:

w,h = 400,250

df = {
  'name':  ['a', 'b', 'c', 'd', 'b'],
  'value': [40,   90, 10,  50,  20 ],
}

p = ggplot(df) + ggsize(w,h)

1. Default presentation¶

1.1. Basic pie chart¶

Use "identity" statistical transformation to leave the data unchanged.

In [4]:

p + geom_pie(aes(slice='value', fill='name'), stat='identity')

Out[4]:

1.2. Improve appearance¶

add stroke (stroke and stroke_color)
make the pie bigger (size)
add hole to draw donut-like chart (hole)
use blank theme (remove axis and grid)
use better colors

In [5]:

p + \
    geom_pie(aes(slice='value', fill='name'), stat='identity',
            size = 20, stroke = 1, stroke_color = 'white', hole = 0.5) + \
    blank_theme + \
    scale_fill_brewer(palette="Set1")

Out[5]:

1.3. Adding labels to pie sectors¶

Let's label the sectors with their names - configure annotations via layer_labels() function:

In [6]:

p + \
    geom_pie(aes(slice='value', fill='name'), stat='identity',
            size = 20, stroke = 1, stroke_color = 'white', hole = 0.5,
            labels = layer_labels().line('@name').size(16)) + \
    blank_theme + \
    theme(legend_position='none') + \
    scale_fill_brewer(palette="Set1")

Out[6]:

1.4. Use "count2d" statistical transformation¶

geom_pie() uses count2d stat by default. It allows to make a slice sizes proportional to the number of cases in each group (or if the weight aesthetic is supplied, the sum of the weights). Also count2d provides variables for proportion ('..prop..') and proportion in percent ('..proppct..').

Using layer_tooltips() prepare the information for tooltips by adding the variables provided by 'count2d':

In [7]:

tooltip_content = (
    layer_tooltips()
        .line('count|@{..count..} (@{..prop..})')
        .line('total|@{..sum..}')
        .format('..prop..', '.0%')
        .format('..count..', '.1f')
        .format('..sum..', '.1f')
)

Apply 'count2d' to get slices proportional to the number of cases.

In [8]:

p + \
    geom_pie(aes(fill='name'),
             size = 20, stroke = 1, stroke_color = 'white', hole = 0.5,
             labels = layer_labels().line('@name').size(16), 
             tooltips=tooltip_content) + \
    blank_theme + \
    theme(legend_position='none') + \
    scale_fill_brewer(palette="Set1")

Out[8]:

Compute weighted sum instead of simple count with aesthetic weight.

In [9]:

p + \
    geom_pie(aes(fill='name', weight='value'),
             size = 20, stroke = 1, stroke_color = 'white', hole = 0.5,
             labels = layer_labels().line('@name').size(16), 
             tooltips=tooltip_content) + \
    blank_theme + \
    theme(legend_position='none') + \
    scale_fill_brewer(palette="Set1")

Out[9]:

Order sectors by count.

The following ordering rule is used for the pie chart: the first slice goes to the left of 12 o'clock and others go clockwise.

In [10]:

p + \
    geom_pie(aes(fill=as_discrete('name', order_by='..count..'), weight='value'),
             size = 20, stroke = 1, stroke_color = 'white', hole = 0.5,
             labels = layer_labels().line('@name').size(16), 
             tooltips=tooltip_content) + \
    blank_theme + \
    theme(legend_position='none') + \
    scale_fill_brewer(palette="Set1")

Out[10]:

2. Pie size depending on data¶

Make the size of the pie chart dependent on the data: map total count ('..sum..' variable) to the size.

Note that it has its own special representation of the size in the legend.

In [11]:

df2 = {
    'x': [1, 1, 1, 1, 1, 1.5, 1.5,   2,   2,   2 ],
    'y': [1, 1, 1, 1, 1,   2,   2, 1.5, 1.5, 1.5],
    's': [3, 1, 2, 1, 4,   1,   3,   3,   3,  1],
    'n': ['a', 'b', 'a', 'c', 'a',  'a', 'b', 'c', 'a',  'b']
}

ggplot(df2) + \
    geom_pie(aes('x', 'y', fill='n', weight='s', size='..sum..'), hole=0.3, 
             tooltips=tooltip_content) + \
    xlim(0.5,2.5) + ylim(0.5,2.5) 

Out[11]:

Mapping fill and size to the same variable:

In [12]:

ggplot({'n': ["a", "b", "c"], 's': [1, 2, 3]}) + \
    geom_pie(aes(fill='n', slice='s', size='n'), stat="identity") + \
    blank_theme

Out[12]:

3. Explode¶

Use values to explode slices away from their center point, detaching it from the main pie.

In [13]:

length = {
  'name' : ['20-50 km', '50-75 km', '10-20 km', '75-100 km', '3-5 km', '7-10 km', '5-7 km', '>100 km', '2-3 km'],
  'count': [1109, 696, 353, 192, 168, 86, 74, 65, 53],
  'explode': [0, 0, 0, 0.1, 0.1, 0.2, 0.3, 0.4, 0.6]
}

ggplot(length) + \
    geom_pie(aes(fill='name', slice='count', explode='explode'), stat='identity',
            stroke=1, stroke_color='black', size=20) + \
    blank_theme + \
    scale_fill_gradient(low='dark_blue', high='light_green')

Out[13]:

In [14]:

calories = {
    'slice': [35, 25, 25, 15],
    'label': ["Apples", "Bananas", "Cherries", "Dates"],
    'explode': [0.1, 0, 0, 0]
}

p4 = ggplot(calories) + \
    blank_theme + \
    scale_fill_brewer(palette='Set1') + \
    ggsize(w, h)

p_pie = p4 + \
    geom_pie(aes(fill='label', slice='slice', explode='explode'), stat='identity', size=18)

p_donut = p4 + \
    geom_pie(aes(fill='label', slice='slice', explode='explode'), stat='identity', hole=0.8, size=18)

bunch = GGBunch()
bunch.add_plot(p_pie + theme(legend_position='none'), 0, 0)
bunch.add_plot(p_donut, w, 0)
bunch

Out[14]: